Research On Compression Of Remote Sensing Image Based On Region Of Interest

With the rapid development of remote sensing technology and increasing expansion of obtained imagery data, exigent request is brought forward for data compression in finite transmission channel. The traditional algorithms don't work well in these tasks and make obvious distortion of image with high compression. The compression algorithms based on region of interest (ROI) provide effective approaches for solving these problems. Focusing on the compression of remote sensing image based on ROI, this thesis's major work includes:The basic theory of image compression is studied and summarized in this thesis. Especially image coding techniques for ROI and progressive image coding techniques are investigated. All of these give an important guidance to further research.By analyzing normal image segmentation algorithms, a new ROI extraction method based on edge feature is proposed. For the purpose of ROI extraction faced on compression mechanism, this method can divide whole image to multilevel ROIs based on consistency of gradient orientation at the edge, such as edge region, texture region, transition region and placidity region and so on. All of these give a sufficient preparative for further compression coding.ROI coding algorithms based on JPEG2000 as well as some relevant improving algorithms are concluded and analyzed. The effect of different promotion coefficients to image reconstruction is contrasted by experiments, and the effect of upgrade extent of wavelet coefficient to performance of compression is analyzed. Then a ROI coding algorithm based on information entropy is proposed and the whole compression is finished through combining with multilevel division of image region. The experimental results show that our compression algorithm has excellent practical performance.